repositories
A selection of my open-source projects and data science work on GitHub.
Frederick Douglas Pearce
Data Scientist, Earth Researcher, and life-long learner. I love digging in to datasets, figuring out what insights can be gained, and communicating results!
The 14 includes 5 repositories with one star each, which their cards don't show.
GitHub Repositories
codefluent
Open-source VS Code extension and web app that scores how effectively you collaborate with Claude Code, grounded in Anthropic's AI Fluency Research.
sportswear-esg-news-classifier
End-to-end, production ML pipeline (Cloud Run + CI/CD) that classifies ESG news for 50+ sportswear brands using an LLM-as-judge, with ML pre-filters and a daily-updated scorecard.
agentfluent
Local-first agent analytics with prompt diagnostics
claude-code-loop
Reusable Claude Code plugin: a supervised dev-loop engine (one routed backlog issue per invocation, human gates, durable ledger). Extracted from AgentFluent.
claude-code-sessions
Canonical public reference for Claude Code's JSONL session data format — posts, reference docs, and security tooling for working with session data safely.
claude-code-data-collective
Curated, sanitized public corpus of Claude Code session data. Downstream of claude-code-sessions.
e-commerce-fraud-detection
Real-time e-commerce fraud detection: 30 engineered features, an optimized XGBoost classifier, and SHAP explainability tuned to minimize false positives.
predict-housing-prices
Predict housing prices by applying regression kriging to housing attributes, satellite images, etc.
BRFSS-diabetes-2024
This project applies machine learning to predict diabetes risk based on the Behavioral Risk Factor Surveillance System (BRFSS) survey data for 2024, and serves as the midterm project for the DataTalksClub Machine-Learning-Zoomcamp course.
509_Report_Analysis
Repo to download 509 reports, extract relevant data from PDFs, and analyze/visualize the data for trends, stats, etc
complex_numbers_clock
This project creates images and video of an analogue clock to use as an intro to complex numbers
us-presidential-vote-analysis
This project analyzes historical US Presidential Election data to better understand the relationship between the Electoral College Vote results and the Popular Vote results.